Application of the ESTARFM algorithm for fusing Sentinel-2 and MODIS NDSI series in the eastern Qilian Mountains

Eastern Qilian Mountains, China It is important to monitor snow cover via high-spatiotemporal-resolution remote sensing data. However, there is a trade-off between the high/low spatial resolution and low/high temporal resolution of available sensors. In this study, an enhanced spatiotemporal adaptive reflectance fusion model (ESTARFM) based on the Google Earth Engine (GEE) was used to resolve this limitation. We fused MODIS and Sentinel-2 normalized difference snow index (NDSI) to generate daily 10-m snow cover data for the eastern Qilian Mountains from November 1, 2021, to May 1, 2022. We then evaluated the accuracy of the fusion images and compared the differences in snow cover distributions with different resolutions. Most fusion studies have focused on algorithm development and applied these algorithms to vegetation phenology. Our findings show that using ESTARFM could increase the accuracy of snow cover monitoring in mountainous areas. The snow cover areas in the MODIS and fusion images exhibit similar changes, while the fusion images provide more detailed and reliable snow distributions. Overall, ESTARFM implemented on GEE can generate reliable and accurate NDSI series, which have advantages over traditional singular datasets, and fusion images created from these series are suitable for describing the snow cover variations in regional-scale areas. This study provides valuable insights for improving the simulation of snowmelt runoff in regional hydrological models. • Performed the ESTARFM to fuse MODIS and Sentinel-2 NDSI series on GEE. • Analyzed the ESTARFM's suitability and uncertainty for fusing NDSI in mountainous areas. • Evaluated the accuracy of fusion NDSI for snow mapping.